I'm surprised they only tested latency and not throughput. Path differences should affect available bandwidth.

You're not likely to be able to test throughput without running into the limitation imposed by whatever instance type or service you are using to deliver it.

An individual service effectively has infinite bandwidth relative to the available bandwidth of any of the major providers.

Not that seeing test results for service-level capabilities would not be interesting, but it isn't really a test of the network itself.

It would be nice to have access to the raw performance data, all 30 days, all 1485 tests. This report is missing even the full list of regions tested. The rest of the data is heavily processed.

Surprised to see Azure having straightest routes and one of the lowest latencies - they always seemed to be an csp underdog to me.

Me too... Working with Azure daily, this is surprising. Although I suspect I am experiencing slowdowns on storage rather than network.

Me three - just using their portal (I assume an example of Azure best practices) doesn't lead to tons of confidence (at least 2 years ago)

This slideshow looks kinda incomplete without the story. Is there a presentation video somewhere, or an article?

Video should be posted at http://www.cvent.com/events/nanog-75/custom-116-948222eca583... eventually.

It looks like this presentation was delivered by ThousandEyes engineers based on their cloud networking performance benchmarks. You can learn more and download the report here: https://www.thousandeyes.com/press-releases/2018-public-clou...

Would be great to see global network performance comparisons with smaller providers like Digital Ocean and OVH. Everyone always knocks AWS on their bandwidth pricing, yet the pricing is in line with other major cloud providers and CDNs (GCP, Azure, Fastly). I would love to see what the actual trade-off looks like. Is there really that much of a difference in performance or reliability between them, or is it just a case of oligopoly pricing.

Would also be nice to see some stats for traffic within the same AZ as well. I generally see ping times between 0.1 and 0.2 ms within the same AZ on AWS, would be nice to know what that looks like on Azure and GCP.

One thing i'm still surprised about, there is still no site that allows you to easily compare pricing between all cloud providers (azure, aws, gcp, do)... seems like someone should of built one by now.

Thank you! It doesn't seem to have digitalocean, but still very useful!